Unexpected RigidBody child behaviour in Bullet

[Vercix]

Godot version: 3.2.3 RC3

OS/device including version:
Windows 10 64-bit

Issue description:

There seems to be a bug with regards to physics of RigidBody childs of RigidBodies.

• Seems to be only happening in Bullet.
• I did not have this issue with the 3.2.3 beta. (I forgot which specific commit I was building from)
• RigidBody children connected by a HingeJoints seem to be completly unaffected by physics.

Behaviour in Bullet:

Behaviour in GodotPhysics:

Steps to reproduce:
Have any RigidBody child to a RigidBody node.

Minimal reproduction project:
test.zip

[Vercix]

I also had the following error stack when starting the project. This does not get thrown every time though and I cant reliably reproduce it. So I dont know if it is related.

ERROR: shape_get_data: Condition "!shape->is_configured()" is true. Returned: Variant()
At: servers/physics/physics_server_sw.cpp:125
ERROR: CylinderShape is not supported in GodotPhysics. Please switch to Bullet in the Project Settings.
At: servers/physics/physics_server_sw.cpp:73
ERROR: get: Condition "!p_rid.is_valid()" is true. Returned: __null
At: ./core/rid.h:149
ERROR: shape_set_data: Condition "!shape" is true.
At: servers/physics/physics_server_sw.cpp:103
ERROR: Invalid ID.
At: servers/physics/physics_server_sw.cpp:1400
ERROR: set_data: Condition "!d.has("cell_size")" is true.
At: servers/physics/shape_sw.cpp:1640
ERROR: get: Condition "!p_rid.is_valid()" is true. Returned: __null
At: ./core/rid.h:149
ERROR: body_attach_object_instance_id: Condition "!body" is true.
At: servers/physics/physics_server_sw.cpp:671
ERROR: Invalid ID.
At: servers/physics/physics_server_sw.cpp:1400

Edit: Now that I'm having a closer look at the error, it seems like its complaining about a CylinderShape in GodotPhysics. I didnt have any CylinderShapes at any time in the test project though.

[Vercix]

I looked at into it a bit more. And the 3.2.2 stable pre-compiled build also gives me the same issue. I have been building Godot from source for the last year or so. And none of those builds were giving me this issue.

I am planning on doing a clean build of the 3.2 branch when I have time to do so.

So I am really not sure what's going on. It might just be my system specific. But I cant imagine what it could be.

[madmiraal]

This issue exists in 3.1-stable too. The issue is not present in 3.0.6.

[madmiraal]

Fixed on master with #40184 i.e. it will be fixed in 3.2 with #40788.